Fault-tolerance and on-line maintainability in a process control supervision system

Abstract

Large and dependable process control and supervision systems are a key component of today's integrated computer manufacturing. These systems must offer various levels of quality of service such as scalability, distribution, fault-tolerance and on-line maintainability. This paper presents how the concept of object groups and process groups have been exploited to meet this challenge. Object groups extend the process group functionalities with an object naming service, a trap in the object messaging for sending messages on the network transparently, and a dispatcher to retrieve the object to which the message received on the network is addressed. The process group concept provides consistent distributed data due to the use of ordered reliable multicasts and group membership management.